Never miss a story from Data Center. Subscribe for in-depth analysis and articles.
In the CadenceLIVE Silicon Valley presentation on From Chips to Chillers: Electronics Cooling Through to Sustainability, Sherman Ikemoto, System Sales Group Director for the Cadence DataCenter Design Software and Insight Platform, discusses the sustainable methods for electronics cooling using Cadence tools. Watch the recorded video to learn more about data center performance problems and how Cadence tools can help mitigate them.
Today, there is concern that data centers could become an unsustainable drain on energy resources, proliferating globally. Although they are the backbone of the digital economy, these centers require high amounts of electricity, increasing the complexity of maintaining and expanding them. Despite these complexities, IT systems are expanding rapidly and globally. Data centers could consume 100 times more power in the future than an equivalent-sized office building. Data centers consume about 2- 3% of the global electricity demand, and experts predict that this number might rise between 5% to 20% by 2030. The increasing demand for electricity escalates costs and impacts global warming and sustainability, raising concerns among organizations and environmentalists.
Energy consumption has remained constant over the past few years, despite advancements in technology and practices. As technology continues to grow, consumption has also increased. To ensure a sustainable future, it is crucial that IT equipment becomes more energy-efficient while facilities evolve to prioritize energy efficiency and sustainability.
All data centers lose performance. Here, performance is defined as customer expectations when investing in maintaining and operating a data center. If a customer builds an IT data center of 1MW power, they will want to build an IT system that uses 0 to 1 MW power without any barriers to reaching that 1MW limit. They would also expect high efficiency. In short, a data center is expected to deliver the compute capacity it was designed for at the lowest possible cost.
Surveys conducted by data center industries generate a lot of information. One such survey conducted by the Global Data Center Alliance on 12,000 data center operators revealed that the typical capacity utilization of an enterprise data center is only 56%. In other words, for a 10MW data center, the usage is only 5MW. These figures illustrate that the present data center industry faces a low utilization rate.
For running a 1MW IT load, two 1MW power IT data centers are required, which adds to the total operational and maintenance costs. Accompanying this increasing cost impact is the environmental impact. In addition, these data centers also experience power outages. In the last three years, 60% of operators have experienced IT service outages in their data centers. The cost incurred to these organizations due to these outages adds up to over a million dollars, which is a significant number, and these costs need to be reduced for sustainably using and maintaining these data centers.
The root cause of this outage can be determined by comparing a server's and data centers' life cycles. So, in the design phase of a server, multiple configurations of a server will be tested before finalizing a configuration that meets the performance requirements. For the selected server configuration, the performance values are locked. Now, in the operational lifespan of the server, the physical configuration remains the same, so the consumer is guaranteed to get the pre-defined performance.
The design phase of a data center is the same as a server’s. However, the performance predicted for the design configuration is no longer valid in the operational phase. The consumer of the data center will load the empty data center with their available devices and equipment, introducing loads different from the ones predicted during the design cycle. These design changes in the facility lead to performance issues unless dealt with accordingly.
In the design phase, simulation tools are used to study the performance of the servers and the data centers. However, the data center customers lack these simulation tools during the operational phase, which results in these performance losses. This, in turn, can lead to service outages or low-capacity utilization and high-capacity costs.
The figure above shows the same data center facility run on two IT road maps. One uses the Cadence DataCenter Insight Platform, and the other uses traditional planning techniques. On tracking the resource utilization, which includes space, airflow, power, and cooling, we notice that resource utilization is accelerating faster in the traditional planning technique and hits the high-risk margin.
Although running a perfect data center is impossible, a digital twin can offer better performance management. With a digital twin, a data center can use up to 90% of the power it was designed to deliver. In traditional planning, adding any further equipment will increase the risk factor, which is precisely why data centers with traditional planning run at 50% – 60% of their total load.
The Cadence DataCenter Design Software and Insight Platform comprise multiple tools that enable thermal design from the chip to the chiller. Cadence Celsius EC is an electronics cooling tool that uses computational fluid dynamics to handle the electronics cooling process from chip to system. The DataCenter Design Software and Insight Platform encompasses the DataCenter Design Software (with a choice of 3 packages – Essential, Advanced, or Pro) and the DataCenter Insight Platform for the detailed design of data centers from the system level up to the room. This platform combines multiple tools, such as space planning, power planning, etc., into a single platform or interface to avoid switching between platforms. Moreover, this platform makes it easy for cross-functional teams to collaborate and work towards efficient decision-making on the data center operational design.
In the design and operational phase, the platform has a digital ecosystem that connects the design flows to the operational process. The digital ecosystem is a catalog of any IT or power or cooling infrastructure component available on the market and would go into a data center.
The DataCenter Insight Platform comprising the Asset Twin Module (DataCenter Asset Twin) and the Digital Twin Module (Datacenter Digital Twin), can be easily added to your data center operations existing workflow, adding visibility to the unseen cooling performance and enabling you to avoid risk, increase energy efficiency, and unlock stranded capacity. To learn more about the current performance problems in data centers and how Cadence tools can help mitigate them, watch the video From Chips to Chillers: Electronics Cooling Through to Sustainability.